RTX 4080 is very underwhelming in my HP Z840 Workstation

[Crunchy Dragon]

59 minutes ago, si1enze said:

That's why I was curious about the model of Philips monitor..

I was interested in the HDMI version. Some revisions of HDMI don't support more than 4k60 anyway, so there's no point in trying to chase more than 60 FPS at 4k depending on what connector you're using.

[da na]

1 hour ago, Hinjima said:

Your CPU is bottlenecking the your GPU hard!..  I had a 9900k@5.1ghz and playing with the RTX 4080 was a nightmare, I had to upgrade to a 7700x and saw massive improvement and sometimes over double the FPS in certain games...

In my z840 I have a 72 thread Xeon e5-2697v4 (rather two) in my z840, 8 channel ECC DDR4-2133, CPU clock of 2.9ghz. Horrible setup for gaming, and on my RTX A4500 (same die as 3080) I have great performance. 
Check your temps, might be running a bit toasty. But as aforementioned, make sure your monitor is set above 60hz, and that your display cable supports it

[jessquirl]

Philips UHD 4K 60Hs  model 50PFL5602/F7 A

 

https://www.usa.philips.com/c-p/55PFL5602_F7/5000-series-smart-ultra-hdtv-with-nettv-and-high-dynamic-range

[starsmine]

12 minutes ago, jessquirl said:

Philips UHD 4K 60Hs  model 50PFL5602/F7 A

 

https://www.usa.philips.com/c-p/55PFL5602_F7/5000-series-smart-ultra-hdtv-with-nettv-and-high-dynamic-range

yea the driver on the display is hard locked to 60 fps
 

Quote

 Resolution

Computer formats

• 640 x 480, 60Hz
• 720 x 480, 60Hz
• 800 x 600, 60Hz
• 1024 x 768, 60Hz
• 1280 x 768, 60Hz
• 1280 x 1024, 60Hz
• 1360 x 768, 60Hz
• 1440 x 900, 60 Hz
• 1920 x 1080, 60Hz
• 1680 x 1050, 60Hz
• 1400 x 1050, 60Hz
• 3840 x 2160, 60Hz
• 4096 x 2160, 60Hz

That does not explain why the gpu is failing to produce more frames then the 1080ti imo. because those should be decoupled. but the TV will lock its output to 60hz.

Also weird that it takes true 4K(4096 x 2160) as well as UHD. 

[Brooksie359]

2 hours ago, NobleGamer said:

Things that other comments may not have suggested yet:

 

1. Make sure that the Windows display settings show a refresh rate of more than 60 Hz. I had a Windows update or something change my ~165 Hz setting to 60 Hz.
2. If you can completely rule out a refresh rate cap: Before you scrap your CPU or build, please run benchmarks that monitor CPU usage (MSI afterburner, some game benchmarks like the Tomb Raiders). If it is pegged near 100% at your highest FPS and you aren't happy with that FPS, then sure that's a CPU bottleneck. If its not pegged at 100%, then you may have to check per core utilization, because various games may rely more on a single thread or core, which could cause only 1-2 cores or threads to get pegged.

You need your own data or data from somewhere that proves that your CPU is in fact a bottleneck before scrapping your build.

I have never seen a cpu bottleneck that had the cpu 100% all core load. The major of the time it will be only one thread that is maxed out. You hardly ever see 100% multicore scaling that can actually fully utilize a cpu. 

[Guest]

1 hour ago, Crunchy Dragon said:

I was interested in the HDMI version. Some revisions of HDMI don't support more than 4k60 anyway, so there's no point in trying to chase more than 60 FPS at 4k depending on what connector you're using.

I agree on the HDMI limitations..

[NobleGamer]

7 hours ago, Brooksie359 said:

I have never seen a cpu bottleneck that had the cpu 100% all core load. The major of the time it will be only one thread that is maxed out. You hardly ever see 100% multicore scaling that can actually fully utilize a cpu. 

Yes, a CPU bottleneck doesnt always literally show 100%, which is why my comment also explained to check for high single thread or single core utilization.

 

My favorite example of this is running Fallout 4 on a 3570K processor with a RTX 3080. That regularly got the CPU close enough to 100% for a CPU bottleneck to be clear.

[Brooksie359]

4 hours ago, NobleGamer said:

Yes, a CPU bottleneck doesnt always literally show 100%, which is why my comment also explained to check for high single thread or single core utilization.

 

My favorite example of this is running Fallout 4 on a 3570K processor with a RTX 3080. That regularly got the CPU close enough to 100% for a CPU bottleneck to be clear.

That is probably because that cpu has a super low thread count. Most cpus have more than 4 threads and so you will likely never see them anywhere near full load during gaming. I would expect a xeon to not have all of the threads loaded up but only some of them. 

[Brooksie359]

4 hours ago, NobleGamer said:

Yes, a CPU bottleneck doesnt always literally show 100%, which is why my comment also explained to check for high single thread or single core utilization.

 

My favorite example of this is running Fallout 4 on a 3570K processor with a RTX 3080. That regularly got the CPU close enough to 100% for a CPU bottleneck to be clear.

Also if your gpu isn't at 100% load you have a bottleneck regardless. That is how I look for bottlenecks as it is way easier to see and more consistent. If you are near 100% load most of the time then it's safe to say that your setup is gpu is being fully utilized. 

[NobleGamer]

6 hours ago, Brooksie359 said:

Also if your gpu isn't at 100% load you have a bottleneck regardless.

Sure, unless the game or its engine is unoptimized - Then you can hit a soft FPS limit before the GPU and/or CPU gets near 100% load. BATTLETECH on Unity is my favorite example of this, which I played before & after CPU + GPU upgrade, and it barely boosted FPS at moderate & far distances.

[Brooksie359]

1 hour ago, NobleGamer said:

Sure, unless the game or its engine is unoptimized - Then you can hit a soft FPS limit before the GPU and/or CPU gets near 100% load. BATTLETECH on Unity is my favorite example of this, which I played before & after CPU + GPU upgrade, and it barely boosted FPS at moderate & far distances.

Obviously if you have a frame cap or an engine level frame limitation then not getting 100% wouldn't necessarily mean it's a cpu bottleneck but that is sorta common sense. 

[Morrie Sells Wigs]

On 3/1/2023 at 2:05 AM, jessquirl said:

I use a Phillips 4K UHD monitor. I use an HDMI cable.

Shouldn't you be using Display Port?

[Fendrick]

Crap old platform, time for a new platform and superior CPU.

[starsmine]

1 hour ago, Fendrick said:

Crap old platform, time for a new platform and superior CPU.

The guy is playing at 4k. the CPU is a bottleneck, but its like a 10% bottleneck. 

The fuck is with these forums in general and their incorrect assumptions with CPUs and gaming.

at 4k

With a GTX 1080
YOU ARE GPU BOTTLENECKED

going from a 1080 to a 4080, should raise framerate, at 4k ACCROSS THE BOARD. 
Even IF you bind the the max FPS down. Min and avg will go up. 

That for OP the frame rate went DOWN means something else entirely. 

Now boom go from 1080 to 2080ti

He should still be getting 80fps in farcry 5 one of the tested games. 
But no, he has a stronger GPU, and since he is not 100% CPU bound, he should be getting 90-100fps in farcry 5, not 60. 

AKA, its NOT his CPU. 

[Fendrick]

OK it is magic fairy dust and the flying pigs.

 

Showing 2080 Ti results, nowhere near 4080 level lol.

 

Nvidia's driver overhead is really bad with RTX 3k and 4k cards, maybe consider the facts.

 

 

 

Latest game to show this massively is Hogwarts Legacy.

 

 

[starsmine]

17 minutes ago, Fendrick said:

OK it is magic fairy dust and the flying pigs.

 

Showing 2080 Ti results, nowhere near 4080 level lol.

Explain why a more powerful GPU then the 1080 made his frame rate go DOWN when I just showed that a more powerful GPU then the 1080 should make his frame rate go UP. 
How much is not the question. How strong a cpu bind is not the question. 

You are right a 2080ti is no where near a 4080. and I pointed that out already, That it has no effect to the problem at all. 

That there is a software issue here and people recommend spending hundreds of dollars on a new platform is wrong. 

[Fendrick]

9 minutes ago, starsmine said:

Explain why a more powerful GPU then the 1080 made his frame rate go DOWN when I just showed that a more powerful GPU then the 1080 should make his frame rate go UP. 
How much is not the question. How strong a cpu bind is not the question. 

You are right a 2080ti is no where near a 4080. and I pointed that out already, That it has no effect to the problem at all. 

That there is a software issue here and people recommend spending hundreds of dollars on a new platform is wrong. 

Driver overhead, there is a drop in performance overall on RTX 30 and 40 if your CPU is weak that yes on earlier cards would give a higher FPS, it's nothing new and is why it seems like it is a different problem, with a Radeon GPU the guy will get better FPS due to much lighter driver overhead.

 

It's not every game that displays this disparity but almost all games here are CPU limited meaning the driver overhead issues are greater here.

 

There is no cutting it, it is still a big CPU bottleneck and you need a modern system for RTX 3k and 4K cards if you want their potential.

[Deadpool2onBlu-Ray]

On 2/28/2023 at 7:51 PM, jessquirl said:

HI, I recently bought an RTX 4080 Founders Edition from Nvidia. I have an HP Z840 Workstation I use as my gaming and CAD machine. After installing the RTX 4080 I have had zero improvements on my gaming frame rates compared to my old GTX 1080 Ti. I get a Max of 60 Fps on any game at any setting. I am disgusted with Nvidias Support. They tell me my CPU (Xeon E5-2687W V4) is the problem, yet I read online that my CPU and the RTX 4080 are a great combination. I really want this RTX 4080 to be awesome but it isn’t.  60 Fps isn’t with the $1,200 I just paid for it. I am going to return in since I have no reason to keep it at this point. Could any of you give me any insight on this problem?

 

 

Jesse

Jesse's msinfo32.nfo 4.01 MB · 5 downloads

2-25-23.pdf 2.05 MB · 9 downloads

Bro your CPU is old and slow AF. Also, what PSU are you using?

[starsmine]

5 minutes ago, Fendrick said:

Driver overhead, there is a drop in performance overall on RTX 30 and 40 if your CPU is weak that yes on earlier cards would give a higher FPS, it's nothing new and is why it seems like it is a different problem, with a Radeon GPU the guy will get better FPS due to much lighter driver overhead.

 

It's not every game that displays this disparity but almost all games here are CPU limited meaning the driver overhead issues are greater here.

 

There is no cutting it, it is still a big CPU bottleneck and you need a modern systemfor RTX 3k and 4K cards if you want their potential.

That... that is not how any of this works. 


This is the test with a a 3080.
the i9-9900k is Skylake based, one Tock newer then Haswell/Broadwell. 

[Fendrick]

Just now, starsmine said:

That... that is not how any of this works. 

Hardware unboxed is faking it then? OK.

 

[starsmine]

4 minutes ago, Fendrick said:

Hardware unboxed is faking it then? OK.

 

Find me the hardware unboxed reviews that argue the point you just made at 4k resolutions



Notice how, even with a CPU bind, the frame rates went UP 



Re read the problem OP is having. 

[Fendrick]

7 minutes ago, starsmine said:

That... that is not how any of this works. 


This is the test with a a 3080.
the i9-9900k is Skylake based, one Tock newer then Haswell/Broadwell. 

What are you not understanding? The 9900K is also not fast enough even for a 3080 LOL.

 

FC5 is a VERY CPU limited game, it still is not negating how much driver overheard these newer Nvidia cards are subjected to.

 

The CPU OP has only reaches 3.5ghz, compared to a 9900K the 9900K will be a lot faster in games 4.7ghz to 5.2ghz as they are still clock speed bound and IPC.

 

 

[Fendrick]

1 minute ago, starsmine said:

Find me the hardware unboxed reviews that argue the point you just made at 4k resolutions



Notice how, even with a CPU bind, the frame rates went UP 

 

I already linked them, perhaps read.

 

Also the 13900K is far far far faster, you know this right? In the order of magnitude of around 2x faster which is gigantic.

 

The 3060 is too slow to make a difference on them too.

[starsmine]

1 minute ago, Fendrick said:

What are you not understanding? The 9900K is also not fat enough even for a 3080 LOL.

 

FC5 is a VERY CPU limited game, it still is not negating how much driver overheard these newer Nvidia cards are subjected to.

 

The CPU OP has only reaches 3.5ghz, compared to a 9900K the 9900K will be a lot faster in games as they are still clock speed bound and IPC.

 

 

What are you not understanding that even the most CPU bound game does not bind you to get lower FPS

[Fendrick]

Forget it, prefrontal cortex problems.